Solution Architect · modern web · AI · data · systems
Perth, Western Australia
Portfolio · Perth House Data · LinkedIn · Email
I design and build expressive, production-minded systems across TypeScript, React, Next.js, cloud, AI, data, integration and identity.
The common thread is making complex things tangible. I am interested in the complete path from an ambiguous problem to a dependable product: architecture, data, implementation, interface and explanation.
Perth House Data is a free, public analytical product built from three decades of Perth house-sale records.
It turns a fragmented historical dataset into something ordinary people can search, understand, compare and download—without hiding the underlying evidence behind a subscription or valuation funnel.
| Coverage | 30 years · 330 suburbs · 281k+ recorded house sales |
| Data path | Source CSV → immutable raw archive → canonical property model → cleaned facts → analytical curation |
| Platform | Next.js · Vercel Workflows and Blob · Neon Postgres · MotherDuck |
| Analysis | Rolling medians · annual movement · sales velocity · land/value relationships · bedroom cohorts |
| Output | Public suburb explorer · downloadable CSV · reusable visual snapshots |
Explore the live product → Read the source →
GitHub Motion Graph turns repository, commit, branch and pull-request activity into a force-directed visual system. It is designed as a modern full-stack application rather than a static visualisation.
Next.js 16 · React Server Components · Edge and Node APIs · Neon Postgres · Drizzle ORM · Vercel
Fabric Semantic Model Starter is a React and TypeScript foundation for Microsoft Fabric Apps. It uses live DAX metadata to turn a Power BI semantic model into an explorable interface rather than leaving it trapped behind specialist tooling.
React 19 · TypeScript · Microsoft Fabric Apps · DAX metadata · Config-driven deployment
ModelViz is a browser-based architecture tool for inspecting Power BI and Microsoft Fabric semantic models—including tables, relationships, data-source modes and row-level security.
Next.js · TypeScript · Local TMDL parsing · Interactive architecture diagrams · Security inspection
| Data and analytical products | Platform and automation architecture | Interactive systems |
|---|---|---|
| Models and public-facing tools that make large or untidy datasets trustworthy and legible. | Durable workflows that connect data, documents, decisions, identity and cloud services. | Visual interfaces and simulations that make complex behaviour easier to inspect and understand. |
I work comfortably from architecture through implementation. That usually means TypeScript, Python and SQL, modern web application patterns, Postgres and analytical databases, Microsoft Fabric and Azure, Vercel, serverless orchestration, and carefully scoped use of language models.
The technology matters, but the standard I care about is simpler: is the result trustworthy, explainable and useful?
- Convergent — A sandbox for multi-agent reasoning, coordination and emergent system behaviour.
- OrbitMe — A Newtonian orbital simulator for spatial computing, inspired by The Expanse.
- next-boids-text — Generative typography driven by flocking behaviour and real-time interaction.
- Pixel-to-Voxel Projector — An experimental image-to-spatial reconstruction pipeline.
- Recupare — Secure invoice intelligence using Azure Content Understanding, isolated user data and structured extraction workflows.
- Find the real shape of the problem. Understand the source data, operational constraints and decisions the system must support.
- Build a durable path through it. Separate raw evidence, canonical identity, transformation and presentation so each layer can be tested and changed safely.
- Make the result legible. A technically correct system still fails if people cannot understand, trust or use it.
If you are working on a data-heavy product, an automation problem that has outgrown manual handling, or an ambitious prototype that needs a sound architecture, I would be glad to hear about it.
hypecoding.dev · LinkedIn · rossfarrell7@gmail.com
Building useful systems from difficult inputs.